FGF1 and insulin control lipolysis by convergent pathways

Abstract

Inexorable increases in insulin resistance, lipolysis, and hepatic glucose production (HGP) are hallmarks of type 2 diabetes. Previously, we showed that peripheral delivery of exogenous fibroblast growth factor 1 (FGF1) has robust anti-diabetic effects mediated by the adipose FGF receptor (FGFR) 1. However, its mechanism of action is not known. Here, we report that FGF1 acutely lowers HGP by suppressing adipose lipolysis. On a molecular level, FGF1 inhibits the cAMP-protein kinase A axis by activating phosphodiesterase 4D (PDE4D), which separates it mechanistically from the inhibitory actions of insulin via PDE3B. We identify Ser44 as an FGF1-induced regulatory phosphorylation site in PDE4D that is modulated by the feed-fast cycle. These findings establish the FGF1/PDE4 pathway as an alternate regulator of the adipose-HGP axis and identify FGF1 as an unrecognized regulator of fatty acid homeostasis.

Keywords: FGF1; PDE4; cAMP; hepatic glucose production; insulin; lipolysis; type 2 diabetes.

Figure 1.. FGF1 suppresses adipose lipolysis

A) Ex-vivo lipolysis in gonadal white adipose tissue (gWAT) explants from wildtype (F1WT) and FGF1 KO (F1KO) mice 6 h after refeeding. Data are represented as mean ± SEM (n=4, *p < 0.05). B) Lipolysis in mouse SVF-derived adipocytes measured by the cumulative release of free fatty acids (FFAs) into the media over 4 h. Cells were pretreated with vehicle (PBS) or FGF1 (100 ng/ml) for 10 min prior to the induction of lipolysis with 1 nM isoproterenol (ISO). Data are represented as mean ± SEM (n = 3, *p < 0.05, **p < 0.01). C) Dose response of FGF1-induced suppression of lipolysis in 3T3-L1 adipocytes. Cells were pretreated with indicated doses of FGF1 for 10 min prior to the induction of lipolysis with 100 nM isoproterenol (ISO), and the cumulative release of FFAs over 4h measured. Data are represented as mean ± SEM (n = 3, **p < 0.01, ***p < 0.001). D) Serum FFA levels in overnight fasted adR1WT and adR1KO DIO mice 30 min after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. Data are represented as mean ± SEM (adR1WT vehicle n = 5, FGF1 n=5; adR1KO vehicle n = 4, FGF1 n=4 *p < 0.05). E) Ex-vivo lipolysis in gWAT explants from overnight fasted adR1WT and adR1KO DIO mice 2 h after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. Data are represented as mean ± SEM (n = 5 per group, *p < .05). F) ³H-labeled oleic acid turnover in chow-fed adR1WT and adR1KO mice. Overnight fasted mice were injected with vehicle (PBS) or FGF1 (0.5 mg/kg, s.c.) 6 hours prior to the portal vein infusion of ³H-labeled oleic acid. Plasma radioactivity was measured by scintillation counting and normalized to t=1 min. Fractional oleic acid turnover rate calculated by linear regression of natural log transformed data (adR1WT vehicle n = 5, FGF1 n = 4; adR1KO vehicle n = 4, FGF1 n = 5). Data are represented as mean ± SEM (* p < .05). G) Ad lib fed blood glucose levels in vehicle (PBS) and FGF1 injected (0.5 mg/kg) ob/ob mice with and without co-administration of the ATGL inhibitor atglistatin (120 mg/kg p.o) (Veh, n = 5; FGF1, n = 4; ATGLi, n = 5; ATGLi+FGF1, n = 6). Data are represented as mean ± SEM. (* p < 0.05). H) Western blots of total and S660 phosphorylated HSL (pHSL) in 3T3-L1 adipocytes 10 min after vehicle or FGF1 (100 ng/ml) treatment. Quantification of pHSL-S660 normalized to total HSL is shown in the right panel. Data are represented as mean ± SEM (n = 4, *p < 0.05). I) Western blots of total and S660 phosphorylated HSL (pHSL) in gWAT from chow-fed C57BL/6J mice 30 min after vehicle (PBS), FGF1 (0.5 mg/kg) or insulin (1 U/kg) injection. Quantification of pHSL-S660 normalized to total HSL (right panel). (Veh, FGF1, n = 5; insulin, n = 4). Data are represented as mean ± SEM (**p < 0.01, ***p <0.001). See also Figure S1.

Figure 2.. FGF1 suppresses hepatic glucose production (HGP) in an adipose FGFR1 dependent manner

A) Pyruvate tolerance test (PTT, left panel) and glycerol tolerance test (Glycerol TT, right panel) in overnight fasted ob/ob mice 2 h after vehicle (PBS) or FGF (0.5 mg/kg) injection. Data are represented as mean ± SEM (n = 5 per group; *p < 0.05, **p < 0.01, #p < 0.001). B) PTTs in adR1WT and adR1KO DIO mice, as described in A. Data are represented as mean ± SEM (n = 4 per group, *p < 0.05, **p < 0.01). C) Heatmap of hepatic metabolites in ob/ob mice 2 h after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. DHAP, dihydroxyacetone phosphate; G6P, glucose 6-phosphate; F6P, fructose 6-phosphate; G1P, glucose 1-phosphate; OAA, oxaloacetate; PG, phosphoglycerate; PEP, phosphoenolpyruvate. Vehicle n=5, FGF1 n=6, *p < 0.05. D) Hepatic G6P, F6P and 2-phosphoglycerate (2-PG) levels in HFD-fed adR1WT and adR1KO mice 6 h after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. adR1WT Veh n=8, FGF1=7; adR1KO veh n=6, FGF1=6. Data are represented as mean ± SEM (*p < 0.05). E) Hepatic acetyl-CoA levels (left panel) and pyruvate carboxylase (PC) activity (right panel) normalized by protein content in mice described in (D) (*p < 0.05). F) Basal and glucose-clamped levels of endogenous glucose production (EGP), glucose infusion rate (GIR), and glucose disposal rate (GDR) in ob/ob mice after one week of vehicle (PBS) or FGF1 (0.5 mg/kg q.o.d.) injections, measured during a hyperinsulinemic clamp. Vehicle n=9, FGF1 n=8. Data are represented as mean ± SEM (*p < 0.05). See also Figure S2.

Figure 3.. FGF1 suppression of lipolysis is dependent on PDE4 activity

A) Kinetics of isoproterenol (ISO, 100 nM) induced cAMP levels in 3T3-L1 adipocytes pretreated with vehicle (PBS) or FGF1 (100 ng/mg) for 15 min, as measured by ELISA. Data are represented as mean ± SEM (n = 4, *p < 0.05, **p < 0.01). B) 3T3-L1 adipocyte lipolysis after vehicle (PBS) or FGF1 (100 ng/ml) treatment in the presence or absence of PDE4 inhibitors (roflumilast, 2 µM and cilomulast, 10 µM). PBS or FGF1 was added 15 min prior to isoproterenol (100 nM) stimulation for 4 h. Data are represented as mean ± SEM (n = 4, ***p < 0.001). C) cAMP levels in 3T3-L1 adipocytes, pretreated for 15 min with vehicle (PBS) or FGF1 (100 ng/ml) with or without PDE4 inhibitor (roflumilast, 2 µM), 30 min after isoproterenol treatment (100 nM). Normalized GFP fluorescence from cAMP biosensor. Data are presented as mean ± SEM (n = 12, *p < 0.05). D) Lipolysis in gWAT explants from overnight-fasted DIO mice pretreated with the PDE4 inhibitor (roflumilast, 5 mg/kg p.o.) 1 h prior to vehicle (PBS) or FGF1 (0.5 mg/kg) injection. Mice were sacrificed 2 h later. Data are represented as mean ± SEM (n = 6, *p < 0.05). E) Kinetics of isoproterenol (ISO, 100 nM) induced perilipin-GFP and HSL-mCherry co-localization in 3T3-L1 adipocytes pretreated for 15 min with vehicle (PBS) or FGF1 (100 ng/ml). Effects of PDE4 (roflumilast, 2 µM; middle panel) and PDE3 (cilostamide, 10 µM; right panel) inhibitors on co-localization. Data are represented as mean ± SEM (n = 12, *p < 0.05, **p < 0.01). F) Lipolysis in 3T3-L1 adipocytes infected with an adipose specific AAV (adAAV) expressing GFP or PDE4D3. Data are represented as mean ± SEM (n = 7, ***p < 0.001). G) cAMP levels in 3T3-L1 adipocytes infected with adAAVs expressing GFP or PDE4D3 30 min after isoproterenol treatment (100 nM), measured using the Green-down biosensor. Data are represented as mean ± SEM (n = 14, **p < 0.01). H) Kinetics of isoproterenol-induced perilipin-GFP and HSL-mCherry co-localization in 3T3-L1 adipocytes infected with adAAVs expressing PDE4D3 or control vector without an open reading frame. Data are represented as mean ± SEM (n = 19, **p < 0.01). See also Figure S3.

Figure 4.. FGF1-induced suppression of lipolysis and blood glucose in dependent on PDE4D

A) Blood glucose levels in ad lib fed DIO mice after administration of vehicle (30 % captisol) or the PDE4 inhibitor roflumilast (5 mg/kg p.o.). Data are represented as mean ± SEM (n = 5 per group, ***p < 0.001). B) Serum FFA levels 1 h after the injection of the vehicle or the PDE4 inhibitor in the mice described in (A). C) ad lib fed blood glucose levels in DIO mice injected with vehicle (PBS) or FGF1 (0.5 mg/kg) in the absence (left panel) or presence (right panel) of the PDE4 inhibitor roflumilast (5 mg/kg). Mice were fasted after the 0 h time point. Data are represented as mean ± SEM (Cntrl Veh n=7, Cntrl FGF1 n=6, iPDE4 Veh n=7, iPDE4 FGF1 n=8 per arm, ** p<0.01, ***p < 0.001). D) Basal and isoproterenol-stimulated (ISO, 1 μM) lipolysis in gWAT explants from overnight fasted, chow-fed control and PDE4D KO mice. Data are represented as mean ± SEM (n = 4 per group, **p < 0.01). E) Isoproterenol-induced lipolysis (1 nM) in SVF-derived adipocytes from control and PDE4D KO mouse. Cells were pretreated with vehicle (PBS) or FGF1 (100 ng/ml) for 10 min prior to Isoproterenol addition. Data are represented as mean ± SEM (n = 4 per treatment, *p < 0.05). F) ad lib fed blood glucose levels in control and PDE4D KO DIO mice after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. Mice were fasted after the injection until the 2 h time point, then food was returned. Data are represented as mean ± SEM (n = 6 per group, *p < 0.05, ***p < 0.001). G) ad lib fed blood glucose levels after FGF1 injection (0.5 mg/kg) in PDE4D KO DIO mice 4 weeks after treatment with adAAVs driving the expression of PDE4D3 or GFP. Data are represented as mean ± SEM (n = 7 per group, *p < 0.05). See also Figure S4.

Figure 5.. PDE4D3-S44 phosphorylation is required for the metabolic effects of PDE4D3

A) Representative Western blot among three independent experiments showing the temporal changes in isoproterenol-induced PDE4D phosphorylation in 3T3-L1 adipocytes pretreated for 15 min with vehicle (PBS) or FGF1 (100 ng/ml). Bracket indicates the phosphorylated, slower migrating PDE4D fraction whereas arrow shows the hypo-phosphorylated form. Quantification of the phospho-band to total is shown below. B) Western blots of PDE4D phosphorylation in gWAT from overnight-fasted chow-fed C57BL/6J mice 30 min after vehicle (PBS) or FGF1 (0.5 mg/kg) injection. Quantification of the phospho-band to total is shown on the right. Data are represented as mean ± SEM (n = 4/arm, #p < 0.001). C) Scheme of mouse PDE4D3 domains and known PKA phosphorylation sites. The conservation of phosphorylation sites between mouse, rat, and human PDE4D3 is shown below. D) Lipolysis in 3T3-L1 adipocytes infected with adAAVs expressing GFP (control), PDE4D3 (4D3), PDE4D3 S44A, or PDE4D3 S85A. Data are represented as mean ± SEM (n = 3 per treatment group, ***p < 0.001). E) Representative Western blot from two-independent experiments showing PDE4D3 expression in 3T3-L1 adipocytes infected with adAAVs expressing wildtype, S44A, S85A, or S44A/S85A PDE4D3 30 min after treatment with isoproterenol (1 μM) and the PDE4 inhibitor roflumilast (4 μM). Arrow indicates hypo-phosphorylated, bracket indicates phosphorylated PDE4D3. Quantification of the phospho-band to total is shown below. F) Representative Western blot from three independent experiments showing PDE4D3 levels after 30min isoproterenol treatment of (100 nM) 3T3-L1 adipocytes infected with adAAVs expressing GFP, wildtype PDE4D3 or PDE4D3 S44A with or without 15 min FGF1 pre-treatment (100 ng/ml) (low exposure-upper panel, high exposure-lower panel). Brackets indicates phosphorylated PDE4D3. Quantification of the phospho-band to total is shown below. G) Western blots of S44 phosphorylated (upper panel) and total PDE4D (lower panel) in 3T3-L1 adipocytes infected with adAAV PDE4D3 after indicated treatments (FGF1, 10 min pretreatment at 100 ng/ml; ISO, 100 nM isoproterenol for 30 min; or FGF1 pre-treatment and 30 min ISO treatment. Quantification of pS44/Total PDE4D is shown in S5E. H) Isoproterenol-induced (1 nM) lipolysis in SVF-derived adipocytes from PDE4D KO mouse infected with adAAVs expressing WT or S44A PDE4D3 pretreated with vehicle (PBS) of FGF1 (100 ng/ml) for 10 min. Data are represented as mean ± SEM (n = 4 per treatment, *p < 0.05). I) Western blots showing PDE4D-S44 phosphorylation in 3T3-L1 adipocytes 15 min after vehicle (PBS) or FGF1 (100 ng/ml) treatment. Cells were pretreated with the PI3K inhibitor wortmannin (5μM) or DMSO 30 min before Veh or FGF1 treatment. Quantification of pS44/total PDE4D is shown in S5H. J) ad lib fed blood glucose levels in ob/ob mice injected with adAAVs expressing GFP (n=9), PDE4D3 (n=8) or PDE4D3 S44A (n = 8). Data are represented as mean ± SEM (*p < 0.05, **p < 0.01) K) ad lib fed serum free fatty acid levels of ob/ob mice described in (J). Data are represented as mean ± SEM (n = 8 per arm; **p < 0.01). L) ad lib fed blood glucose levels of the mice described in (J) after FGF1 (0.5 mg/kg) injection. Mice were fasted after the injection until the 4 h time point, then food was returned. Data are represented as mean ± SEM (adAAV GFP and adAAV PDE4D3, n =7; adAAV PDE4D3 S44A, n = 5; ***p < 0.001). M) Western blots of S44 phosphorylated (upper panel) and total (lower panel) PDE4D in overnight-fasted and 4 h refed mice maintained on chow and HFD. Quantification of pS44 levels normalized to total PDE4D levels. Data are represented as mean ± SEM (*p < 0.05, **p< 0.01). See also Figure S5.

Figure 6.. Model of FGF1-induced suppression of lipolysis and HGP

Distinct FGF1 signaling parallels insulin-induced suppression of adipose lipolysis and hepatic glucose production (HGP). FGF1/FGFR1 signaling in adipocytes activates PDE4D to decrease cAMP levels and thereby PKA activity. Reduced PKA activity attenuates HSL phosphorylation/translocation to suppress lipolysis. FGF1-induced suppression of lipolysis reduces hepatic glucose production (HGP) through the allosteric regulation of pyruvate carboxylase.